2 AI Stocks to Avoid (Including BigBear.ai) and 1 to Buy Now

The article argues that BigBear.ai and C3.ai are losing relevance, citing deteriorating fundamentals and weak outlooks: BigBear.ai revenue fell from $146 million to $128 million from 2021 to 2025, while C3.ai’s revenue is projected to decline from $389 million in fiscal 2025 to $251 million by fiscal 2028. Broadcom is positioned as the preferred AI beneficiary, with AI chip sales up 65% to $20 billion in fiscal 2025 and expected to reach $60 billion-$90 billion by fiscal 2027. Overall, the piece is negative on smaller AI software names and constructive on Broadcom.

Analysis

The market is increasingly separating AI infrastructure monetizers from AI wrappers. The weak names here are not just losing share; they are getting trapped in a negative operating-leverage loop where lower growth forces heavier reliance on services, restructuring, or acquisitions that further dilute margins. That matters because in crowded AI software, the winner is often the vendor with the lowest integration friction and the strongest distribution, not the one with the best model narrative.

Second-order, Broadcom’s ASIC exposure is a direct beneficiary of hyperscalers’ desire to diversify away from GPU concentration risk. The more Meta/Alphabet-style custom silicon programs scale, the more the AI capex pie shifts from merchant accelerators toward design wins, networking, and adjacent infrastructure software — a subtle headwind for pure-play GPU sentiment but a tailwind for names embedded in the buildout. Broadcom’s mix also gives it a cleaner path to pricing power than software peers whose AI features are becoming table stakes.

The bearish view on BBAI and AI is likely still underestimating how long investor patience can remain absent earnings inflection. These can stay optically cheap for months, but until they show durable recurring revenue quality, any rally is likely to be multiple expansion rather than fundamental re-rating. The real risk to the short is a takeover, contract lumpiness, or a liquidity-driven squeeze, not operating improvement.

Consensus is probably not fully appreciating the asymmetry in AVGO: if AI silicon demand merely meets the low end of current expectations, the stock is supported by cash-flow durability; if custom ASIC adoption accelerates, there is another leg of estimate revisions. The bigger downside risk is not execution failure but valuation compression if AI capex growth slows before Broadcom’s non-AI businesses re-accelerate.

Key Decisions for Investors

• Go long AVGO on any 3-5% pullback over the next 2-6 weeks; thesis is estimate durability plus upside optionality from ASIC design wins. Favor call spreads over outright stock if implied vol remains elevated.
• Initiate a pairs trade: long AVGO / short BBAI or AI, sized 1:1 notional, to isolate quality-of-revenue and margin-compression dispersion over 3-9 months. Best risk/reward if the tape turns risk-off and weak balance sheets are punished.
• For tactical downside in BBAI, use puts rather than stock short if borrow is tight; target a 1-2 month horizon into any sector rotation. The trade works if growth disappoints or the market stops paying for acquisition-driven revenue.
• Avoid initiating new longs in AI unless there is evidence of subscription mix recovery and restructuring stabilization; otherwise use rallies to fade. The stock may remain range-bound, but upside is capped until gross margin inflects.
• If holding NVDA or other AI beta, consider partial hedge with a small AVGO overweight: it captures AI spend while reducing reliance on merchant GPU multiple expansion.